On 02/08/2025 21:33, Ihor Radchenko wrote:
P.S. I am confused why multiple people are asking to disable cache. Org
has been creating ltximg directory since forever and nobody ever asked
to "disable" that one.

Ihor, do not try to find greatest common denominator that is suitable for everybody. A part of users prefer to have directories with documents clean and do not mind to have some stuff in ~/.cache (anyway it is used by many other applications directly or through libraries like fontconfig). Others believe that it is better to store data more locally in respect to documents. Maybe other categories exist as well. We may try to discuss what are common use cases and what heuristics may be used to guess what kind of user started Emacs.

You managed to alleviate pain with slow navigation and agenda refresh. However enough users have never faced that trouble. Is it possible to refrain from writing caches if a file is small enough or if it was instantly parsed? Maybe it is possible to ask user in a non-distracting way when some slow operations are detected.

The following post suggests that many Emacs packages may write to .emacs.d:

Re: How to disable completely org-persist. Sun, 1 Jan 2023 16:42:36 +0300.
<https://list.orgmode.org/Y7GNzJUyhzls7Y6S@protected.localdomain>

Perhaps some Emacs-global configuration may be introduced and users should not adjust configuration of every file. Just an idea

((directories .
  ((xdg_cache . t)
   (emacs_d . nil)
   (tmpdir . nil)))
 (packages .
  ((org . t)
   (t . nil)))) ; disabled for other packages

Maybe more variants should be allowed: ask, force-disabled...

Org is not the only source of annoyance. Emacs developers decided that .eln cache must be namely per user. I have several system users to keep files related to different roles isolated. Org is installed as a *system* package. Why .eln files can not be created once and I have to hear compiling for every user instead? Please, do not post recipes related to eln cache in response. I wrote it just to illustrate effect of the decision.

P.S. I admit, browsers use caches rather extensively and controlling it is not really trivial.

Reply via email to